Distributed ensemble based iterative classification for churn analysis and prediction of dropout ratio in e-learning

Kumaran, V. Senthil; Malar, B.

doi:10.1080/10494820.2021.1956547

Cited by 2 publications

(1 citation statement)

References 16 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The results show that Ensemble Learning has a marginal increase in classification performance with the best accuracy value being 69.22% with a passing precision of 56.8%, a dropping out precision of 77.44%, a passing recall of 62.49% and a dropping out recall of 73.04% [13]. In addition, Ensemble Learning on the Naive Bayes method, SVM and Logistic Regression has also been used for grouping dropout students showing that Ensemble can improve predictive performance significantly with 94.24% accuracy, 95.62% precision, 93.12% recall, 94.1% f-measure and 0.91% AUC [14]. Individual models may have high bias (underfitting) or high variance (overfitting), but by aggregating their predictions, ensemble methods can achieve a better balance between the two.…”

Section: Introductionmentioning

confidence: 99%

An Ensemble Voting Approach for Dropout Student Classification Using Decision Tree C4.5, K-Nearest Neighbor and Backpropagation

Cholis,

Ulinnuha

2023

IJAIDM

View full text Add to dashboard Cite

Many factors cause drop out in students. This study classified active students and drop out students using 1092 student data consisting of 557 active student data and 535 drop out student data. The independent variables used are Semester, Semester Credit Units (SKS), Semester Grade Point Average (IPS), Grade Point Average (IPK), admission pathways and Single Tuition Fee (UKT). Classification is carried out using the Ensemble Voting method where the method will combine the Decision Tree C4.5, KNN and Backpropagation methods as a single method. In addition to knowing the classification of active students and drop out students, this study aims to prove whether the Ensemble Voting method is able to get better results than the single method. This classification using a comparison of training and testing data of 90:10 to build model. Classification results from a single method will be included in the Ensemble Voting method. The Decision Tree C4.5 method gets 95.45% accuracy, 98.03% precision and 92.59% recall. KNN gets 96.36% accuracy, 100% precision and 92.59% recall. Backpropagation gets 90.90% accuracy, 95.83% precision and 95.18% recall. Meanwhile, the Ensemble Voting rule used is Ensemble Soft Voting with a weight of (2,1,1). Ensemble Voting with Ensemble Soft Voting rules is able to improve the accuracy, precision and recall values with 98.18% accuracy, 100% precision and 96.29% recall.

show abstract

Section: Introductionmentioning

confidence: 99%